JPH05122677A - Encoder and decoder for video signal - Google Patents

Abstract

PURPOSE:To obtain an efficient encoder by dividing a band with the frequency component in the diagonal direction of a picture and performing the encoding according to each frequency band. CONSTITUTION:In the encoder the sample point of the staggered grid from the sample point in a square grid shape is derived by interpolating a video signal inputted from an input terminal 50 with an interpolation circuit 400. By making the video signal in the grid square shape pass through the first diagonal division filter 410 in the two diagonal direction to filters 450 and 460 respectively, the encodings 500 to 530 are performed after band-dividing the video signal in the diagonal grid shape at the two-dimensional frequency. In the decoder the encoded video signal 60 is used as inputs 70 and the decoder restores the sample point in the staggered grid shape while synthesizing the signal at every frequency band by passing filters 800 and 810 respectively in the two diagonal directions after encoding 710 to 740 at every frequency divided on the encoder. Further it restores an original signal by thinning out the data of the sample point obtained by the interpolation at the initial stage on the encoder.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video signal coding apparatus and a video signal decoding apparatus used for digitizing and transmitting or recording a video signal.

[0002]

2. Description of the Related Art In recent years, high quality television images have been actively promoted, and among them, a high-definition television system which is completely different from the current television system has been proposed by the Japan Broadcasting Corporation (see, for example, "References"). Special issue: High-definition television ”, Journal of Television Society, Vol. 36, No. 1, 1
982). This high-definition television system has about five times as much information as the current television system.
Therefore, one of the important issues is how to efficiently transmit and record such a high-definition television signal having a large amount of information, and various high-efficiency coding systems have been proposed.

One of the conventionally proposed effective coding systems is a sub-band coding system (for example, "pre-sub-band intra-sodium prediction coding system" Proceedings of National Conference of Television Society, PP457-PP458). , 1989
Year). This sub-band coding method is based on a quadrature-mira filter (QMF) proposed by Johnston (J, D, Johnston), etc.
for Use in Quadrature Mi
error Filter Banks ") in PRO
C. 1980 IS-SP (ICASS
P) PP291 to PP294) is used to divide a video signal into different frequency bands and optimal coding and quantization are performed for each frequency band.

FIG. 6 is a block diagram showing the structure of a conventional subband coding system. In FIG. 6, a video signal is input from an input terminal 10, passes through a horizontal division filter (H division filter) 100 that performs band division at a horizontal frequency, and then a vertical division filter (that performs band division at a vertical frequency). V division filter) 110,
Encoders 150 and 16 that are input to 120, are divided, and perform encoding in each of the divided frequency bands.
The coded video signal, which is added to the multiplexing circuit 200 through 0, 170 and 180, is output from the output terminal 20. The above is the configuration of the encoding device.

In the figure, the encoded video signal is input to the input terminal 30 and, after passing through the demultiplexing circuit 300, decoders 310, 320 and 330 for decoding each frequency band. 340.
Further, the video is decoded after being synthesized with horizontal frequency components by a horizontal synthesis filter (H synthesis filter) 370 through vertical synthesis filters (V synthesis filters) 350 and 360 for synthesizing frequency components in the vertical direction. The signal is output from the output terminal 40. The above is the configuration of the decoding device.

The operation of the conventional coding apparatus and decoding apparatus configured as described above will be described below with reference to FIGS. 6, 7, and 8. Here, FIG. 7 corresponds to FIG.
8 is a sample pattern of the encoding apparatus shown in FIG. 8, and FIG. 8 is a diagram showing characteristics of band division in FIG. Further, the horizontal division filter 100, the horizontal synthesis filter 370, and the vertical division filters 110 and 12 shown in FIG.
0 and the vertical synthesis filters 350 and 360 are respectively
It is composed of the above-mentioned quadrature-mira-fitter (QMF).

Hereinafter, the encoding device will be described first.
In FIG. 6, the video signal input from the input terminal 10 of the device is sent to the horizontal division filter 100 and divided into a low frequency component (L) and a high frequency component (H) at the horizontal frequency. Low frequency component (L) and high frequency component (H) of the divided horizontal frequency
Are sent to the vertical division filters 110 and 120, respectively. The vertical division filters 110 and 120 divide the low frequency components and the high frequency components at the vertical frequency, and the encoders 150 and 16
Send to 0, 170, 180. Here, the transition of sample points and the state of band division will be described with reference to FIGS. 7 and 8.

The horizontal division filter 100 and the vertical division filters 110 and 120 shown in FIG. 6 are composed of QMF or the like, and at the time of band division, sample points are thinned out in half in the filter direction. FIG. 7 shows the transition of the sample points at this time. The grid-shaped sample points of the original image shown in FIG.
00, the number of pixels is halved in the horizontal direction as shown in FIG.
As a result, the number of pixels is further halved in the vertical direction as shown in FIG. 6 (c), so that the sample points after band division are ¼ of the original image. Further, FIG. 8 shows a state of band division at this time on a two-dimensional frequency. Since band division is performed by a horizontal filter and a vertical filter, as shown in FIG. It is divided by a straight line parallel to the horizontal frequency axis and the vertical frequency axis.

A description will be given below with reference to FIG. Encoder 1
The video signals sent to 50, 160, 170 and 180 are
Encoding is performed here. Encoders 150, 160, 1
Reference numerals 70 and 180 each include a circuit including an encoding circuit and a quantization circuit, and are encoded with encoding characteristics and quantization characteristics suitable for each frequency band. Multiplexing circuit 200
Then, the video signals encoded by the encoders 150, 160, 170 and 180 are temporally switched to perform multiplexing. The encoding device is configured by the above configuration, and the encoded video signal is output from the output terminal 20.

Next, the decoding device will be described. In FIG. 6, the video signal input from the input terminal 30 and encoded is sent to the demultiplexing circuit 300. Demultiplexing circuit 3
In 00, the signals for each frequency that are temporally multiplexed on the encoding side are separated, and the decoders 310, 320, 330,
Sent to 340. Decoders 310, 320, 330,
340 is a block that performs processing reverse to that of the encoder on the encoding side, and is configured by a circuit including a decoding circuit and a dequantization circuit, and according to each encoding method performed on the encoding side, Decoding is performed for each frequency band.

The vertical synthesizing filters 350 and 360 frequency-synthesize the video signals decoded by the decoders 310, 320, 330 and 340 for each frequency band at vertical frequencies. After this, the horizontal synthesis filter 370
Then, the original video signal is restored by frequency-wise synthesis at the horizontal frequency and output from the output terminal 40. The decoding device is configured by the above configuration.

[0012]

In the conventional sub-band coding system configured as described above, the video signal is divided into a plurality of frequency bands at the horizontal frequency and the vertical frequency, and is appropriately selected according to each frequency band. It has been proposed in the past that the circuit configuration is simple and the coding efficiency is good. However, in such a conventional configuration, since the band division is performed by the horizontal frequency and the vertical frequency, the pattern of the band division on the two-dimensional frequency becomes a square lattice, and the high frequency components in the diagonal direction are visually detected. However, even though it is a component that is not very noticeable, the same encoding as the low frequency component is performed for both the horizontal frequency and the vertical frequency, which are important for the image, and there is a problem that sufficient encoding efficiency cannot be obtained. Was.

The present invention is to solve the above-mentioned problems, in which the band-splitting pattern is formed in an oblique lattice pattern on a two-dimensional frequency,
For the diagonally invisible frequency components,
An object of the present invention is to provide a coding device and a decoding device with high coding efficiency by performing rough coding of a quantization step similar to a high frequency component of a horizontal frequency or a vertical frequency up to a relatively low component. To do.

[0014]

In order to solve the above problems, in the coding apparatus of the present invention, first, the sample points at the fifth grid position are obtained by interpolation from the square grid sample points and the sample points are divided into five. The video signal is band-divided into a diagonal grid pattern on a two-dimensional frequency by sequentially filtering the sample points of the fifth grid pattern from two diagonal directions in order to perform coding. It was done. Further, in the decoding device of the present invention, the video signal encoded as described above is input, and after decoding for each frequency band divided on the encoding side, the filters are sequentially filtered from two diagonal directions. The signal of each frequency band is synthesized by passing through it to restore the fifth-point grid sample points, and the original signal is obtained by thinning out the sample point data obtained by interpolation at the first stage on the encoding side. It was designed to restore.

[0015]

According to the present invention, the sampling points of the square grid are first interpolated to obtain the sample points at the position of the five-eye grid, and the sample points of the five-eye grid are formed. The video signal is band-divided at diagonal frequencies by sequentially filtering from the first to the third.The diagonal components that are visually unnoticeable are similar to the high-frequency components of the horizontal and vertical frequencies up to relatively low components. Coarse coding of the quantization step can be performed, and the coding efficiency can be increased.

[0016]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An encoding apparatus and a decoding apparatus according to an embodiment of the present invention will be described below with reference to the drawings. Figure 1
FIG. 1 is a block diagram showing the configurations of an encoding device and a decoding device according to an embodiment of the present invention.

In FIG. 1, a video signal is input from an input terminal 50 and sent to a five-eye lattice interpolation circuit 400 which interpolates a sample point at a fifth-eye lattice position from a square lattice sample point. After that, the signal passes through a first diagonal division filter 410 that performs band division at a first diagonal frequency, and further into second diagonal division filters 450 and 460 that divide at a second diagonal frequency different from the first direction. Sent. Next, it passes through encoders 500, 510, 520, and 530 for encoding the band-divided video signals, and a multiplexing circuit 600 that temporally multiplexes the signals encoded by the respective encoders. Sent. The encoded video signal is output from the output terminal 60. The above is the configuration of the encoding apparatus according to the present embodiment.

Further, the video signal encoded in the figure is input from the input terminal 70 and sent to the demultiplexing circuit 700, and the temporally multiplexed signal is demultiplexed. After that, the decoders 710, 720, 730 and 740 perform decoding for each frequency band. Next, it passes through second diagonal synthesis filters 800 and 810 which synthesize again the frequency bands divided by the second diagonal division filter on the encoding side, and further,
The frequency bands divided by the first diagonal band division filter on the encoding side are sent to the first diagonal synthesis filter 850 which synthesizes again. Finally, in the fifth-eye lattice thinning circuit 900,
The original sampling points of the square lattice are restored and output from the output terminal 80 by thinning out the sampling points at the position of the fifth-eye lattice first performed on the encoding side.

The operation of the encoding apparatus and the decoding apparatus of the present embodiment configured as described above will be described below with reference to FIGS. 1, 2 and 3. 2 is shown in FIG.
FIG. 4 is a pattern diagram of transitions of sample points of the encoding device and the decoding device shown in FIG. 3, and FIG. 3 is a diagram showing band division in FIG. 1 on a two-dimensional frequency.
Further, in FIG. 1, the first diagonal division filter 410 and the first
Diagonal synthesis filter 850, second diagonal division filter 45
0,460 and the second diagonal synthesis filters 800,810
Filters are applied in the same direction on each sample pattern to divide or combine the bands, and is composed of the quadrature mirror filter (QMF) described above.

The operation of the coding apparatus of this embodiment will be described below. In FIG. 1, the input terminal 5 of the device
The video signal input from 0 is a five-eye grid interpolation circuit 40.
The sample point at the fifth grid position within the square grid sample point of 0 is obtained by interpolation from the peripheral sample points,
The sample pattern is converted into a five-eye grid. Next, in the first diagonal division filter 410, the band division in the diagonal direction is carried out by passing the sample points of the fifth-eye grid pattern through the filter in the diagonal direction. Similarly, in the second diagonal division filters 450 and 460, band division is performed through the filters in the diagonal directions different from the first diagonal division filter for the sample points of the respective five-eye grids.

Now, the transition of the sampling points will be described with reference to FIG. First, the square-lattice-shaped original sample points shown in FIG.
Are converted to sample points in a five-eye grid pattern as shown in FIG. Next, the first diagonal division filter 4
In the 10th and second diagonally divided filters 450 and 460, the sampling points are thinned to half in the filtering process direction based on the above-mentioned QMF principle, and the sample points are shown in FIG. FIG. 3 shows the band division at this time on a two-dimensional frequency. In the frequency region of the original image shown in FIG. 3A, the band in the diagonal direction is not changed by the five-eye lattice interpolation circuit. , The horizontal band and the vertical band are doubled to become the band shown by the dotted line in FIG. Next, the band division shown in FIG. 3C is obtained by filtering from two different diagonal directions.

Hereinafter, the description of the coding apparatus of this embodiment will be continued with reference to FIG. Second diagonal division filters 450, 46
The signals band-divided by 0 are coded by the encoders 500 and 510 in accordance with the respective frequency bands and sent to the multiplexing circuit 600. In the multiplexing circuit 600, the coded signals of the four systems are multiplexed on the time axis and output from the output terminal 60 of the device. With the above configuration, the encoding apparatus according to the present embodiment is configured, and the output terminal 60 outputs the encoded video signal for each band divided by the diagonal frequency.

Next, the decoding apparatus of this embodiment is shown in FIG.
Will be explained. In FIG. 1, the encoded video signal input from the input terminal 70 of is the demultiplexing circuit 70.
Sent to 0. The demultiplexing circuit 700 demultiplexes the signals of the respective frequency bands temporally multiplexed on the encoding side and sends them to the decoders 710, 720, 730, 740. Decoders 710, 720, 730 and 740 respectively decode the signals coded for each frequency band and send them to second diagonal synthesis filters 800 and 810. The second diagonal synthesis filters 800 and 810 correspond to the second diagonal division filters 450 and 460 on the encoding side, and the signals divided by the second diagonal frequency on the encoding side are synthesized again here. The one-diagonal synthesis filter 850 is sent.

The first diagonal synthesis filter 850 corresponds to the first diagonal division filter 410 on the coding side, and the signal divided on the coding side at the first diagonal frequency is synthesized again here. , Five-eye lattice thinning circuit 90
Sent to 0. The fifth-eye lattice thinning-out circuit 900 corresponds to the fifth-eye lattice interpolation circuit 400 on the encoding side,
On the encoding side, the sample points at the positions of the fifth-eye grid, which are first interpolated, are thinned out to restore the original sample points on the square grid and output from the output terminal 80 of the apparatus. Here, the transition of the sample points will be described again with reference to FIG.

First, the sample points of the sample pattern shown in FIG. 2D are filtered from two different directions by the second diagonal synthesis filters 800 and 810 and the first diagonal synthesis filter 850 shown in FIG. Figure 2
The sample pattern shown in (e) is obtained. After that, the five-point grid decimation circuit 900 shown in FIG. 1 thins out the sample points at the positions of the five-eye grid to obtain a sample pattern shown in FIG. 2 (f), and the original pattern shown in FIG. 2 (a). Image sample points are restored. The decoding apparatus according to the present embodiment is configured with the above configuration, and the original video signal decoded for each frequency band is restored and output from the output terminal 80 of the apparatus.

In the above embodiment, the band division from two different diagonal directions is explained as a single stage, but this band division may be recursively used for the components having low horizontal frequency and vertical frequency of the image. The operation when band division is performed with such a configuration will be described with reference to FIGS.
Show.

FIG. 4 is a diagram showing the transition of the sample pattern when such band division is performed in two consecutive steps, and FIG. 5 shows the two-dimensional frequency of the band division pattern at this time. It is shown in. The case where the band division according to the present embodiment is recursively followed will be described below with reference to FIGS. 4 and 5. The square-lattice-shaped sample pattern shown in FIG.
The sample pattern has a five-eye grid pattern as shown in (b).
Further, the diagonal band division of the second stage results in a square grid sample pattern shown in FIG. FIG. 5 shows band division on the two-dimensional frequency at this time. In FIG. 5, the outer square region is the band component of the original video signal, the rhombic region is the low band component when the first diagonal band division is performed, and the inner square region is the second stage. The low-frequency component when diagonal band division is performed is shown.

Further, in the above-mentioned configuration, the sample pattern in the shape of a square lattice is input, the sample pattern is converted to the form of a five-eye lattice by interpolation from the peripheral sample points, and then the filter processing is performed. When the process of converting analog data to digital data is included, the sample pattern at the time of analog / digital conversion is used as line offset sampling in advance, and is divided into bands in the diagonal direction as a sample pattern in the form of a five-point grid. You may go. The video signal coding device in this case receives an analog video signal as an input, and when sampling the analog video signal, performs offset sampling for each line to obtain a five-point grid sample point. Analog / digital conversion means,
The configuration is provided with diagonal band division means for performing band division from two different diagonal directions with respect to the five-point grid sample points from the analog / digital conversion means, and
A video signal decoding device receives as input a video signal coded by band-splitting in two different diagonal directions from five-eye grid sample points, and inputs two different five-eye grid sample points. An oblique band synthesizing means for synthesizing the bands in the oblique direction to interpolate the sample points by a factor of 2 and a digital / analog converting means for converting the sample points of the fifth-eye grid pattern into an analog video signal are provided. The digital / analog converting means obtains an analog video signal for the output signal interpolated by the oblique band synthesizing means.

[0029]

As is apparent from the above embodiments, in the encoding apparatus and the decoding apparatus according to the present invention, the sample pattern is formed into an oblique lattice shape and then band division is performed through an oblique filter. The signal is divided into frequency components in the diagonal direction. For the frequency components in the diagonal direction, coarse coding of the quantization step is performed up to a relatively low range, and for the frequency components in the horizontal direction and the vertical direction, Since it is possible to perform fine encoding in the quantization step up to high frequencies, it is possible to reduce the number of quantization bits for encoding high-frequency components in the diagonal direction that are visually inconspicuous and perform efficient encoding. It can be carried out.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configurations of an encoding device and a decoding device according to an embodiment of the present invention.

FIG. 2 is a pattern diagram showing transition of sample patterns of the encoding device and the decoding device.

FIG. 3 is a pattern diagram showing frequency division of the encoding device and the decoding device.

FIG. 4 is a pattern diagram showing transition of sample patterns according to another embodiment of the present invention.

5 is a pattern diagram showing frequency division corresponding to FIG.

FIG. 6 is a block diagram showing the configurations of a conventional encoding device and decoding device.

FIG. 7 is a pattern diagram showing the states of sample patterns of a conventional encoder and decoder.

FIG. 8 is a pattern diagram showing frequency division of a conventional encoding device and decoding device.

[Explanation of symbols]

 400 Five-eye lattice interpolation circuit 410 First diagonal division filter 450 Second diagonal division filter 460 Second diagonal division filter 800 Second diagonal synthesis filter 810 Second diagonal synthesis filter 850 First diagonal synthesis filter 900 Five-eye lattice thinning circuit

Claims (6)

[Claims] 1. A video signal sampled in a square lattice shape on a two-dimensional plane is input, and the position of a five-eye lattice in the square lattice is obtained by interpolation from surrounding sample points, and a five-eye lattice is obtained. Different for the five-eye lattice interpolating means for obtaining the sample points of the five-eye lattice and the five-eye lattice-like sample points from the five-eye lattice interpolating means.
A video signal coding apparatus, comprising: diagonal band division means for performing band division from two diagonal directions. 2. A video signal encoded by band-division into two diagonal directions different from a five-point grid sample point is input, and the video signal is sampled from two different diagonal directions with respect to the five-grid sample point. Is performed, and the diagonal band synthesizing means for interpolating the sample points in each of the two different diagonal directions by a factor of two, and the sample points at the fifth eye lattice position are thinned out from the fifth eye lattice-like sample points. By this means, it is provided with five-eye lattice thinning-out means for obtaining square-lattice-like sampling points, and the output signal interpolated by the oblique band synthesizing means is square-lattice-like sampling points using the five-eye lattice thinning-out means An apparatus for decoding a video signal, characterized in that 3. The video signal coding apparatus according to claim 1, wherein the coding is performed after band division is performed recursively using the fifth-eye lattice interpolation means and the diagonal band division means. 4. The video signal decoding device according to claim 2, wherein the sampling points of the original image are obtained by recursively using the oblique band synthesizing means and the fifth-eye lattice thinning-out means. .. 5. An analog / digital conversion means for inputting an analog video signal and performing offset sampling for each line when the analog video signal is sampled to obtain a five-point grid sample point. 2. An oblique band dividing means for dividing the band from two different oblique directions with respect to the sample points of the five-eye grid from the analog / digital converting means.
An encoding device for the video signal described. 6. A video signal coded by band-division into two diagonal directions different from a five-point grid sample point is input, and the video signal is divided from two different diagonal directions with respect to the five-grid sample point. Diagonal band synthesizing means for performing double band interpolation on each sample point and digital / analog converting means for converting from the five-point grid sample points to an analog video signal. 3. The video signal decoding apparatus according to claim 2, wherein an analog video signal is obtained by the digital / analog conversion means for the output signal interpolated by the band synthesizing means.